Description: The research was aimed at developing a deep learning based models for driver activity recognition capable of utilizing spatial as well as temporal dimensions of the data. The novel contribution was inclusion of Hijabi and veiled drivers in the dataset to addresss the underrepresentation of female drivers in the exisiting literature. The model employed achieved the accuracy upto 97.99%.
Socio-Economic Benefits: Improvement of roadside management as using this approach can help in preventing the accidents
Category: Computing
Department: CSSE